In television market analysis, it is important to estimate how many viewers have seen a particular program that is presented in a number of different local markets. For example, such information can be used to set advertising rates and to determine local network ratings. In another application, determining whether particular programs are shown can be used to ascertain if local content presenters are presenting programs in accordance with required program schedules imposed by networks or other program sources.
One common way of estimating how many viewers have potentially seen a program is to record what shows were seen by a small sample of viewers in a particular market. The data representative of viewing by the small sample of viewers is then extrapolated to predict the viewing behavior of a larger audience. Alternatively, digital watermarks or other codes can be embedded in programs and detected by listening receivers in each of the markets. Both methods may be inaccurate depending on how many viewers are in the sample set, how well the data is recorded, or how well the watermarks are detected.
Given these problems, there is a need for an improved system for estimating the number of viewers that have seen a program and that can reliably determine if a particular program was presented within a local market.